Jetson Orin NX 16GB: 48 benchmark rows, 25% measured.

Method v1.0 · dataset 2026-09-07 · verified 2026-09-07 · last updated September 2026

12 of 48 rows are measured and published (class C); the rest are EdgeAIStack-derived (class D). 14 model families. Precisions: FP32 (4), FP16 (29), INT8 (15). Runtimes: tensorrt (48). Tasks: detection (39), classification (4), lidar detection 3d (1), speech recognition (2), pose (1), segmentation (1). Jetson Orin NX 16GB platform page.

Family × precision: best value

The fastest matching row per family and precision (fps, or tokens/s for language models), with its evidence class. Click a cell for the full breakdown.

Family	FP32	FP16	INT8
EfficientDet	—	—	248 fps D
MobileNetV2	—	1250 fps D	—
MobileNetV3	—	1420 fps D	—
PointPillars	—	14 fps D	—
ResNet	—	1180 fps D	1681.9 fps C
RetinaNet	—	21 fps C	—
RT-DETR	—	46.5 fps D	61.9 fps D
Whisper	—	— D	—
YOLO11	95 fps D	211 fps C	256 fps C
YOLO12	—	178.7 fps D	218 fps D
YOLOv8	—	193 fps D	264 fps D
YOLOv8	—	175 fps D	—
YOLOv8	—	160 fps D	—
YOLOv9	—	195 fps D	—

Query this hardware live.

Filter by model, precision or runtime and get the matching rows with a permanent link.

OPEN BENCHMARK EXPLORER →

Jetson Orin NX 16GB: 48 benchmark rows, 25% measured.

Family × precision: best value

Every model on Jetson Orin NX 16GB

Links

Query this hardware live.